This package is a yarn workspace referenced by mlflow/server/js/package.json as @mlflow/eslint-plugin. In mlflow/server/js/.eslintrc.js, we automatically load the recommended rules in index.js.

Adding New Custom Rules

Create a new .js file in this directory (e.g., my-custom-rule.js)
Export an ESLint rule module with meta and create functions:

module.exports = {
  meta: {
    type: 'problem',
    docs: {
      description: 'Description of your rule',
      category: 'Best Practices',
      recommended: true,
    },
    messages: {
      myMessage: 'Your error message here',
    },
  },
  create(context) {
    return {
      // Your rule implementation
    };
  },
};

Import the rule in index.js, and add it to the rules sections of the export (add to the top-level rules config as well as the one in recommended):

const myCustomRule = require('./my-custom-rule');

module.exports = {
  rules: {
    // ...
    'my-custom-rule': myCustomRule,
  },
  configs: {
    recommended: {
      plugins: ['@mlflow'],
      rules: {
        '@mlflow/my-custom-rule': 'error',
      },
    },
  },
};

Run yarn install from mlflow/server/js to make sure the new rules are loaded
Run yarn lint from mlflow/server/js to make sure the new rule works as intended

Testing Custom Rules

Run yarn test to run all tests in the directory. We currently use ESlint's RuleTester util to write unit tests for our rules. Alternatively you can spot-check by running yarn lint in mlflow/server/js on files you expect to fail your rule.